

% All variables are available in the sheet 'data'. 
% There is the description of the variables in the sheet 'Description'. 
% The variable ECBpre in the paper was constructed using Detailed data on Eurosystem purchases that 
% are proprietary and available only to their researchers. We signed a protocol in which
% we had to delete the variable and the corresponding raw data after running all the needed statistical analyses.
% We have recreated a pseudo-variable that replicates the same moments presented in Table 2.
% We do not expect all the other numbers to be exactly the same, but we do expect a similar qualitative conclusion.
% We provide a pdf with the similar tables as in the paper with this variable

% This code allows to replicate the main results in the paper, corresponding to Tables 2, 3, 4, 5, 6 and 7 in the paper.


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%% TABLE 2 Summary statistics of main variables
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
tabstat  RBAS100 DRIFT100 SIZE COVER SPREAD VOL ME SE TE GE SDUR,  s(n mean sd min p25 p50 p75 max) format(%5.2f) col(stat)
% just ECB row
tabstat  ECBpre_fake if Auction>6,  s(n mean sd min p25 p50 p75 max) format(%5.2f) col(stat)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%% TABLE 3 Correlations
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
corrtex  RBAS DRIFT SIZE COVER SPREAD VOL ME SE GE TE SDUR, file(correlation) replace case sig dig(2)
% just ECB row
% This second correlation matrix excludes the first six auctions in the data that were conducted during a 	period where the ECB was not yet active in the secondary market
% Only the last row is of relevance for the results in Table 3 in the paper (see the last row in the table in the paper)
corrtex  RBAS DRIFT SIZE COVER SPREAD VOL ME SE GE TE SDUR ECBpre_fake if Auction>6, file(correlation) replace case sig dig(2)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%% Table 4 Determinants of the marginal elasticity
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
eststo clear
quietly reg ME RBASn DRIFTn SIZEn SPREADn VOLn ECBpren_fake dummyECB SDUR , robust 
eststo model1 
quietly reg ME RBASn DRIFTn SIZEn SPREADn VOLn ECBpren_fake dummyECB SDUR i.Date_yyyy i.dq , robust
eststo model2 
quietly reg ME RBASn DRIFTn SIZEn SPREADn ECBpren_fake dummyECB SDUR, robust
eststo model3 
quietly reg ME RBASn DRIFTn SIZEn SPREADn ECBpren_fake dummyECB SDUR i.Date_yyyy i.dq, robust
eststo model4 
esttab, r2 ar2 t scalar(rmse) coeflabels(RBASn RBAS DRIFTn DRIFT SIZEn SIZE COVERn COVER SPREADn SPREAD VOLn VOL SEn SE ECBpren_fake ECBpre_fake dummyECB D\_ECB) b(3) t(3) r2(3) star(* 0.10 ** 0.05 *** 0.01) stats(N r2_a)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% Table 5 Predictive regressions of the 5-day ahead abnormal return
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
eststo clear
quietly reg AR5 RBASn DRIFTn, robust
eststo model1
quietly reg AR5 RBASn DRIFTn SIZEn COVERn SPREADn VOLn SEn, robust
eststo model2
quietly reg AR5 MEn, robust
eststo model3
quietly reg AR5 MEn RBASn DRIFTn, robust
eststo model4
quietly reg AR5 MEn RBASn DRIFTn SIZEn COVERn SPREADn VOLn SEn, robust
eststo model5
esttab , replace nonumbers coeflabels(MEn ME RBASn RBAS DRIFTn DRIFT SIZEn SIZE COVERn COVER SPREADn SPREAD VOLn VOL SEn SE SDUR_MEn SDUR_ME) order(MEn)  b(2) t(2) r2(3)    ar2 star(* 0.10 ** 0.05 *** 0.01) label title(Predictive Regression) stats(N r2_a )
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%% Table 6 Predictive regressions at various holding horizons with all controls
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
eststo clear
forval i=1/10 {
	quietly reg AR`i' MEn RBASn DRIFTn SIZEn COVERn SPREADn VOLn SEn, robust
	eststo model`i' 
	}
esttab , replace nonumbers coeflabels(MEn ME RBASn RBAS DRIFTn DRIFT SIZEn SIZE COVERn COVER SPREADn SPREAD VOLn VOL SEn SE SDUR_MEn SDUR_ME) order(MEn)  b(2) t(2) r2(3)    ar2 star(* 0.10 ** 0.05 *** 0.01) label title(Predictive Regression) stats(N r2_a )
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%% Table 7 Cross-sectional test with bond duration
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
eststo clear
forval i=1/10 {
	quietly reg AR`i' MEn RBASn DRIFTn SIZEn COVERn SPREADn VOLn SEn SDUR SDUR_MEn, robust
	eststo model`i' 
	}
esttab , replace nonumbers coeflabels(MEn ME RBASn RBAS DRIFTn DRIFT SIZEn SIZE COVERn COVER SPREADn SPREAD VOLn VOL SEn SE SDUR_MEn SDUR_ME) order(MEn)  b(2) t(2) r2(3)    ar2 star(* 0.10 ** 0.05 *** 0.01) label title(Predictive Regression) stats(N r2_a )
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
